A senior International Monetary Fund official has identified a hidden risk in the artificial intelligence boom: debt markets. Tobias Adrian, the IMF's financial counselor, said AI leverage is more worrying than elevated stock valuations, pointing to potential financial stability concerns embedded in corporate debt tied to the sector.
The warning shifts focus from frothy equity markets to the less visible build-up of borrowing among AI-focused firms. As companies race to fund massive compute infrastructure and model development, the accumulation of debt could create systemic vulnerabilities if interest rates rise or revenue projections falter.
In a separate development, OpenAI said it fixed an issue where Codex, its AI coding agent, was burning through user credits faster than normal. Thibault Sottiaux, the engineering lead, explained the tool was performing more backend computation than intended, consuming more compute per task. The company reset usage limits for all affected users after convening a Sunday "warroom."
Users had reported hitting limits more quickly for similar tasks compared to the prior week. Codex usage limits display as a percentage on dashboards and vary by subscription tier. The glitch underscores how AI companies still grapple with opaque compute costs that directly impact user experience.
Debt dynamics and operational hiccups both highlight the growing pains of an industry scaling at breakneck speed. Yet the IMF's concern carries more weight for systemic risk, as even a contained AI sector debt shock could ripple through credit markets.